Publicação
Diffusion-Weighted Imaging in Breast Magnetic Resonance
| Resumo: | Breast cancer is the most frequent, prevalent and mortal cancer affecting women. An earlier and more accurate diagnosis may change this scenario. Different methods are being explored to improve MRI diagnosis of this disease, namely through Diffusion-Weighted Imaging (DWI) and its different diffusion models. The diffusion models studied in this thesis were: the monoexponential, IntraVoxel Incoherent Motion (IVIM), Diffusion Kurtosis Imaging (DKI), IVIM+DKI, stretched exponential, truncated or statistical, and Gamma Distribution (GD). This work aimed to characterize different groups of breast lesions, establish differences among these groups using diffusion models, and compare their diagnostic performances. Additionally, this thesis aimed to study the optimal b-value combination in the DKI model for usage in clinical practice. Women with breast lesions were scanned with MRI and an additional diffusion-weighted sequence was acquired. Lesions were classified in types and subgroups through histology, and the corresponding diffusion models parameters were obtained. Statistical analysis investigated the differences of these parameters and their diagnostic performances were assessed. For the study of optimal b-value combination, the available b-values were exhaustively combined and tested in terms of diagnostic performance for the DKI model. In this optimization study, some principles were depicted and should be considered in DKI studies to minimize the DWI standardization issues. The GD and the statistical model were applied to breast lesions for the first time in this thesis, showing the capability to characterize and to significantly differentiate groups of lesions. The results showed that it is possible to characterize breast lesions using DWI in a robust way, with Gaussian and non-Gaussian diffusion models. These diffusion models also provided differentiation among different groups of lesions. Some non-Gaussian diffusion models surpassed the performance of the monoexponential model for breast cancer diagnosis. This work strongly supports the DWI use to improve the MRI role in breast cancer diagnosis. |
|---|---|
| Autores principais: | Borlinhas, Filipa |
| Assunto: | Breast lesions Magnetic Resonance Imaging (MRI) Diffusion-Weighted Imaging (DWI) diffusion models b-values Lesões mamárias Ressonância Magnética (RM) imagem ponderada em difusão (DWI) modelos de difusão valores de b. |
| Ano: | 2019 |
| País: | Portugal |
| Tipo de documento: | tese de doutoramento |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| Resumo: | Breast cancer is the most frequent, prevalent and mortal cancer affecting women. An earlier and more accurate diagnosis may change this scenario. Different methods are being explored to improve MRI diagnosis of this disease, namely through Diffusion-Weighted Imaging (DWI) and its different diffusion models. The diffusion models studied in this thesis were: the monoexponential, IntraVoxel Incoherent Motion (IVIM), Diffusion Kurtosis Imaging (DKI), IVIM+DKI, stretched exponential, truncated or statistical, and Gamma Distribution (GD). This work aimed to characterize different groups of breast lesions, establish differences among these groups using diffusion models, and compare their diagnostic performances. Additionally, this thesis aimed to study the optimal b-value combination in the DKI model for usage in clinical practice. Women with breast lesions were scanned with MRI and an additional diffusion-weighted sequence was acquired. Lesions were classified in types and subgroups through histology, and the corresponding diffusion models parameters were obtained. Statistical analysis investigated the differences of these parameters and their diagnostic performances were assessed. For the study of optimal b-value combination, the available b-values were exhaustively combined and tested in terms of diagnostic performance for the DKI model. In this optimization study, some principles were depicted and should be considered in DKI studies to minimize the DWI standardization issues. The GD and the statistical model were applied to breast lesions for the first time in this thesis, showing the capability to characterize and to significantly differentiate groups of lesions. The results showed that it is possible to characterize breast lesions using DWI in a robust way, with Gaussian and non-Gaussian diffusion models. These diffusion models also provided differentiation among different groups of lesions. Some non-Gaussian diffusion models surpassed the performance of the monoexponential model for breast cancer diagnosis. This work strongly supports the DWI use to improve the MRI role in breast cancer diagnosis. |
|---|